Interpretable Machine Learning Models to Predict the Resistance of Breast Cancer Patients to Doxorubicin from Their microRNA Profiles

Abstract

Doxorubicin is a common treatment for breast cancer. However, not all patients respond to this drug, which sometimes causes life-threatening side effects. Accurately anticipating doxorubicin-resistant patients would therefore permit to spare them this risk while considering alternative treatments without delay. Stratifying patients based on molecular markers in their pretreatment tumors is a promising approach to advance toward this ambitious goal, but single-gene gene markers such as HER2 expression have not shown to be sufficiently predictive. The recent availability of matched doxorubicin-response and diverse molecular profiles across breast cancer patients permits now analysis at a much larger scale. 16 machine learning algorithms and 8 molecular profiles are systematically evaluated on the same cohort of patients. Only 2 of the 128 resulting models are substantially predictive, showing that they can be easily missed by a standard-scale analysis. The best model is classification and regression tree (CART) nonlinearly combining 4 selected miRNA isoforms to predict doxorubicin response (median Matthew correlation coefficient (MCC) and area under the curve (AUC) of 0.56 and 0.80, respectively). By contrast, HER2 expression is significantly less predictive (median MCC and AUC of 0.14 and 0.57, respectively). As the predictive accuracy of this CART model increases with larger training sets, its update with future data should result in even better accuracy.

A Comprehensive Investigation of Active Learning Strategies for Conducting Anti-Cancer Drug Screening

“…Ensemble learning has been used commonly for improving the prediction accuracy by combining prediction results generated by models trained using different data partitions [3] and/or feature subsets/modalities [56,57]. The results generated from the multiple models within the ensemble are fused [58][59][60] using voting mechanisms, such as simply taking the average, to produce final prediction outcomes [58,59,61]. In this work, the uncertainty and mean prediction values estimated from the ensemble models help identify candidate cancer cell lines to obtain response measurements.…”

“…Active learning strategies may also be extended to cancer patient data [60,62,63], possibly with the assistance of transfer learning [3]. Via transfer learning, a model pretrained on cell line drug response data can be used in the active learning procedure with patient data.…”

FINDER: A Fluidly Confined CRISPR‐Based DNA Reporter on Living Cell Membranes for Rapid and Sensitive Cancer Cell Identification

“…[10] However, molecular markers are usually shared by similar cell subpopulations, and their abundances are low and rapidly change. [11] Therefore, a sensitive and rapid identification method based on multiple markers is a promising approach for accurate identification of cancer cells. [12] Aptamer can recognize and modulate cancer cell membrane surface-specific proteins.…”

A Boolean-based machine learning framework identifies predictive biomarkers of HSP90-targeted therapy response in prostate cancer

“…Yet, the ‘curse of dimensionality’ widespread in pharmacogenomics data - where the number of molecular features often far exceeds the number of biological samples - necessitates the development of feature selection strategies for ML algorithms ( Huang et al, 2018 ; Nguyen et al, 2021 ; Ogunleye et al, 2022 .). For example, Ballester and others have developed a scheme termed Optimal Model Complexity (OMC) aimed at identifying a smaller subset of informative features from the much larger original feature space, and integrated OMC with various ML algorithms ( Bomane et al, 2019 ; Naulaerts et al, 2020 ; Nguyen et al, 2021 ).…”